The Exchanges

Q I mean, didn't like Uber like suck or Waymo suck up the entire CMU robotics department at one point?

A Yeah, exactly. So entire, that's, and that's happened multiple times. These companies like May Mobility who basically are, are the U of M lab. You have a Voyage who came out of, uh, Udacity. And so, yeah, that's happening left and right. If you take that mobile analogy though, And then you think, well, in 2010, there's 11 to 12 excitement that just a few years later, 2015, 2016, nobody's, you know, writing objective C, these waves go really, really fast, and I think a good kind of, ah, adage, I think it was Bill Gates who said this, you know, in two years, nothing looks different, but every 10 years, things are dramatically different. So if we look back at 2017, the autonomy doesn't look that much different, the players are generally the same, but I think 10 years from now, autonomy will be very, very different, insofar as it might even be a commodity.

AI assessment note: “Yeah, exactly. So entire, that's, and that's happened multiple times.”

Q Well, since we're talking about right now, and I agree that we don't know what we don't know, who are the players in the ecosystem right now? Like I can guess some of the obvious ones, like the manufacturers of cars, the Mapping companies, the mobilize that, you know, supply components and sensors. Like how would you break down the taxonomy of the players?

A I think the automotive industry is a good analog to some degree, uh, of, of what I think the autonomy industry will be. You'll have end consumer facing companies who will have brands that interface with the consumer. Now, whether those are ride sharing companies, AV providers, or the continue to be the BMW or the Teslas, uh, I think that that's, that's up for debate. Then you'll have folks who are Supplying services. Right now in the, in automotive business services, quote unquote, are the dealer services. But in the autonomy world, we always talk about and is the emergence of the software car. And so in the software car, those services are much more, they look like kind of your phone. I think that seems fairly obvious because you see some of those already. CarPlay and Android Auto are early indications of that. Uh, and then you have the thing that you can call the infrastructure companies, uh, just like you have in phones and in the web. There's this, you know, every time you go to San Jose, you see these office parks of companies you've never heard of, and you wonder, uh, why do they have 10 glass buildings?

AI assessment note: “You'll have end consumer facing companies... Then you'll have folks who are Supplying services.”

Q I mean, didn't like Uber like suck or Waymo suck up the entire CMU robotics department at one point?

A Yeah, exactly. So entire, that's, and that's happened multiple times. These companies like May Mobility who basically are, are the U of M lab. You have a Voyage who came out of, uh, Udacity. And so, yeah, that's happening left and right. If you take that mobile analogy though, And then you think, well, in 2010, there's 11 to 12 excitement that just a few years later, 2015, 2016, nobody's, you know, writing objective C, these waves go really, really fast, and I think a good kind of, ah, adage, I think it was Bill Gates who said this, you know, in two years, nothing looks different, but every 10 years, things are dramatically different. So if we look back at 2017, the autonomy doesn't look that much different, the players are generally the same, but I think 10 years from now, autonomy will be very, very different, insofar as it might even be a commodity.

AI assessment note: “Yeah, exactly. So entire, that's, and that's happened multiple times.”

Q I mean, one could argue that that, to your point, in terms of defining what the iPhone moment is, is not the moment, it's actually the experience of the iPhone. It's the applications, the, uh, The iPhone phenomenon, even. So that's really what we're talking about here. So then on that front, where do you guys think we are? Um, how far away are we from there?

A So we have shuttles already. This is another, I think, mis, mis characterization or classification of autonomy. It's almost always exclusively thought as robo-taxis. And autonomy is actually much more that, the, the adage that anything that moves will one day be autonomous. We believe that very, very deeply. And so the, the point being is they'll come in these like little waves and, and each of those are different. The robo-taxi wave is kind of a bit orthogonal. To the shuttles wave, which is a real thing, which is, uh, campus shuttles, uh, retirement communities. Uh, so those are different, which is orthogonal to the self-driving truck wave, which is orthogonal to the, I would say the, the warehouse robots.

AI assessment note: “So we have shuttles already... they'll come in these like little waves”

Q Right. I, uh, let's go deeper into the differences between digital and physical AI, and more so into Where, where are we today? What progress has been made? What are some of the main major bottlenecks in physical AI? Why don't you unpack some of that?

A Yeah, I mean, I think a lot of times people think about the progress in physical AIs limited to basically two case use cases, and they're just because they're obvious and interesting, which is robotaxes and humanoids. Um, they're very visceral, they're, they're, they excite you, and they're kind of sci-fi. Um, I think they're, those are very interesting. Uh, and they, there is real work being done by us and other people in, in those domains. I think all the other, uh, All the other domains I think are going to be just as important. I mean, you just, you just think about what happens on a port. That's, there's a huge unlock there. Um, and that's, I think that's the, that's the area we're, we're, we're really focused on. It's like all the other nooks and crannies. If you look at like, we, we've talked before about, uh, the rise of Cisco and how, you know, networking kind of went from, you know, first individual machines and companies would get network and then entire countries were getting network. There's a similar thing happening with AI. AI is Getting to that level of kind of sovereign AI is now a discussion. Sovereign AI really is about physical AI, because that's where you're talking about AI in defense, you're talking about AI in the physical machines that are moving around. If you look just at the example of, of, uh, Waymo from America and Pony from China trying to deploy …

AI assessment note: “people think about the progress in physical AIs limited to basically two case use cases”

Q these companies, you know, Uber, WhatsApp, Snap, you know, Airbnb, et cetera, that emerged in quick succession. And so now that technology is getting there for the infrastructure for physical AI, what are some use cases or companies that you could, obviously it's hard to predict the future, but what were you most excited for? Like, what, what could we be talking about the equivalent here of in quick succession?

A I mean, I think Uh, you know, midterm, we want Dana, if not the short term, to really, you know, make humanoids way more real. Uh, there's, I mean, how many, it's like a thousand core tasks in a home from, uh, from humanoids. And these companies, it's like such, I mean, if you talk to people who work in these companies, it's, everything is difficult. Every step of the way is difficult. Collecting data is difficult. Uh, you know, cleaning that data is difficult. Training those models or deploying the model is difficult. And the bar being, I want a high school kid to make a humanoid. So that, that's our, our, our path, and we think there, there could be a lot there, but that's like these, the obvious stuff. I think the true non-obvious stuff is gonna be, we'll look back, we'll be, we'll be way, way more interesting.

AI assessment note: “to really, you know, make humanoids way more real”

Q I mean, one could argue that that, to your point, in terms of defining what the iPhone moment is, is not the moment, it's actually the experience of the iPhone. It's the applications, the, uh, The iPhone phenomenon, even. So that's really what we're talking about here. So then on that front, where do you guys think we are? Um, how far away are we from there?

A So we have shuttles already. This is another, I think, mis, mis characterization or classification of autonomy. It's almost always exclusively thought as robo-taxis. And autonomy is actually much more that, the, the adage that anything that moves will one day be autonomous. We believe that very, very deeply. And so the, the point being is they'll come in these like little waves and, and each of those are different. The robo-taxi wave is kind of a bit orthogonal. To the shuttles wave, which is a real thing, which is, uh, campus shuttles, uh, retirement communities. Uh, so those are different, which is orthogonal to the self-driving truck wave, which is orthogonal to the, I would say the, the warehouse robots.

AI assessment note: “they'll come in these like little waves and, and each of those are different”

Q in this podcast about this importance of differentiation, if this is a tool that everyone has, it sounds like they would differentiate on data. So how do you in this ecosystem where you're making this argument that there's this horizontal versus vertical layer, are all these players willing to share in the ecosystem, the mapping companies, the sensor companies, the big vehicle companies, how do you navigate the data side?

A Data means a lot of different things. It's not like scenarios and, you know, the data that you have for autonomy, but it is The autonomy engineer who themselves are understanding how are the methodologies to best develop an autonomy system. There is some, what we call light network effects there between companies. Well, I mean, if you go to Stanford, they teach classes that help you learn ANSYS's simulation tools. So there's literally this public company called ANSYS that does simulation tools and you can learn how to use it by taking classes at Stanford. And, and that's the same thing with AutoCAD. If you look back, if you look there, there are many tools that kind of fall into the, into this group. I mean, uh, when you learn how to program, you're actually just learning tools. Now what's happened with software development is those tools have become a really just a commodity and there, and there's many different ways. And so we're still in a, quite a nascent niche field with autonomy. So the tools are not a commodity. These tools are so hard to build. These two, the simulation is, it's not a trivial thing to build.

AI assessment note: “Data means a lot of different things. It's not like scenarios”